Add changes for variable epssm and reference sounding option #1131
Conversation
…fied profile 2) Added a logical switch (reference_sounding) to specify whether the thermodynamic variables are computed as a perturbation from a specified reference sounding. If false, the full thermodynamic variables are used and the pressure gradients on the large time steps are computed using log(p).
1) Code added to ../src/core_atmosphere/mpas_atm_core.F to compute vertical profiles of epssm that vary with height at the start of the time integration. These profiles are determined by the parameters config_min_coeff, config_max_coeff, config_transition_lower_bound, and config_transition_upper_bound, and are specified in namelist.atmosphere. 2) Code added to allow thermodynamic variables to be represented either as perurbations from a specified reference state or as the full variables without a reference state. This option is controlled by the logical config_reference_sounding as specified in namelist.init_atmosphere. When configured using the full variables, the horizontal and vertical pressure gradients computed on the large time steps are written in terms of log(p).
There was a problem hiding this comment.
In src/core_atmosphere/Registry.xml
config_min_coeff -> config_epssm_minimum
config_max_coeff -> config_epssm_maximum
config_transition_lower_bound -> config_epssm_transition_bottom_z
units=“-“ -> units=“m”
config_transition_upper_bound -> config_epssm_transition_top_z
units=“-“ -> units=“m”
(Also - code changes to handle the namelist variable name changes)
Change descriptions in Registry
<var name="etp" type="real" dimensions="nVertLevels" units="unitless"
description="etp = (1+epssm(z))/2 at theta point”/>
<var name="etm” type="real" dimensions="nVertLevels" units="unitless"
description="etm = (1-epssm(z))/2 at theta point”/>
<var name="ewp" type="real" dimensions="nVertLevelsP1” units="unitless"
description="ewp = (1+epssm(z))/2 at w point”/>
<var name="ewm” type="real" dimensions="nVertLevelsP1” units="unitless"
-
description="ewm = (1-epssm(z))/2 at w point”/>
In src/core_atmosphere/mpas_atm_core.F
Let’s delete the following commented lines
! min_coeff = 0.1_RKIND
! max_coeff = 0.5_RKIND
! transition_lower_bound = 8000._RKIND
! transition_upper_bound = 13000._RKIND
! call mpas_log_write('test $i $r', intARgs=(/k/),realArgs=(/rdzw(k)/))
In src/core_atmosphere/dynamics/mpas_atm_time_integration.F
Let’s delete my (WCS_epssm) old code in subroutine atm_srk3:
!----------------------
!SK_epssm: setting the 'mesh' pool pointer
call mpas_pool_get_subpool(domain % blocklist % structs, 'mesh', mesh)
!----------------------------
!------------------------
!SK_epssm: added this to retrieve the pointer for rdzw
call mpas_pool_get_array(mesh, 'rdzw', rdzw)
!------------------
! WCS_epssm
! allocate storage for epssm column arrays and set values
! allocate(ewp(nVertLevels+1))
! allocate(ewm(nVertLevels+1))
! allocate(etp(nVertLevels))
! allocate(etm(nVertLevels))
! call set_epssm(ewp,ewm,etp,etm,nVertLevels,rdzw)
! WCS_epssm
! deallocate(ewp)
! deallocate(ewm)
! deallocate(etp)
! deallocate(etm)
In subroutine atm_compute_vert_imp_coefs_work
Let’s get rid of the test print
! temporary print to check etm and ewm
integer :: iprint=0
if(iprint.eq.0) then
do k=1,nVertLevels
call mpas_log_write(' k, etp(k), ewp(k) = $i $r $r', intArgs=(/k/), realArgs=(/etp(k),ewp(k)/))
end do
iprint = 1
end if
Let’s get rid of old WCS code that commented out
! WCS_epssm
! cofwr(k,iCell) =.5dtsepsgravity*(fzm(k)*zz(k,iCell)+fzp(k)*zz(k-1,iCell))
! WCS_epssm
! cofwz(k,iCell) = dtseps*c2*(fzm(k)*zz(k,iCell)+fzp(k)*zz(k-1,iCell)) &
! *rdzu(k)*cqw(k,iCell)*(fzm(k)*p (k,iCell)+fzp(k)*p (k-1,iCell))
! coftz(k,iCell) = dtseps* (fzm(k)*t (k,iCell)+fzp(k)*t (k-1,iCell))
! WCS_epssm
! cofwt(k,iCell) = .5*dtseps*rcv*zz(k,iCell)*gravity*rb(k,iCell)/(1.+qtotal) &
! *p(k,iCell)/((rtb(k,iCell)+rt(k,iCell))*pb(k,iCell))
! WiCS_epssm
! b_tri(k) = 1. &
! +cofwz(k ,iCell)*(coftz(k ,iCell)*rdzw(k )*zz(k ,iCell) &
! +coftz(k ,iCell)*rdzw(k-1)*zz(k-1,iCell)) &
! -coftz(k ,iCell)*(cofwt(k ,iCell)*rdzw(k ) &
! -cofwt(k-1,iCell)*rdzw(k-1)) &
! +cofwr(k, iCell)*(cofrz(k )-cofrz(k-1))
! WCS_epssm
! alpha_tri(k,iCell) = 1./(b_tri(k)-a_tri(k,iCell)*gamma_tri(k-1,iCell))
! gamma_tri(k,iCell) = c_tri(k)*alpha_tri(k,iCell)
In subroutine atm_advance_acoustic_step_work
Let’s get rid of the print statement
! temporary print to check etm and ewm
integer :: iprint=0
if(iprint.eq.0) then
do k=1,nVertLevels
call mpas_log_write(' k, etm(k), ewm(k) = $i $r $r', intArgs=(/k/), realArgs=(/etm(k),ewm(k)/))
end do
iprint = 1
end if
Let’s get rid of the old commented out WCS code
! WCS_epssm
! rs(k) = rho_pp(k,iCell) + dts*tend_rho(k,iCell) + rs(k) &
! - cofrz(k)*resm*(rw_p(k+1,iCell)-rw_p(k,iCell))
! WCS_epssm
! ts(k) = rtheta_pp(k,iCell) + dts*tend_rt(k,iCell) + ts(k) &
! - resm*rdzw(k)*( coftz(k+1,iCell)*rw_p(k+1,iCell) &
! -coftz(k,iCell)*rw_p(k,iCell))
! WCS_epssm
! wwavg(k,iCell) = wwavg(k,iCell) + 0.5*(1.0-epssm)*rw_p(k,iCell)
! WCS_epssm
! rw_p(k,iCell) = rw_p(k,iCell) + dts*tend_rw(k,iCell) &
! - cofwz(k,iCell)*((zz(k ,iCell)*ts(k) &
! -zz(k-1,iCell)*ts(k-1)) &
! +resm*(zz(k ,iCell)*rtheta_pp(k ,iCell) &
! -zz(k-1,iCell)*rtheta_pp(k-1,iCell))) &
! - cofwr(k,iCell)*((rs(k)+rs(k-1)) &
! +resm*(rho_pp(k,iCell)+rho_pp(k-1,iCell))) &
! + cofwt(k ,iCell)*(ts(k )+resm*rtheta_pp(k ,iCell)) &
! + cofwt(k-1,iCell)*(ts(k-1)+resm*rtheta_pp(k-1,iCell))
! WCS_epssm
! rw_p(k,iCell) = (rw_p(k,iCell)-a_tri(k,iCell)*rw_p(k-1,iCell))*alpha_tri(k,iCell)
! WCS_epssm
! wwAvg(k,iCell) = wwAvg(k,iCell) + 0.5*(1.0+epssm)*rw_p(k,iCell)
! WCS_epssm
! rho_pp(k,iCell) = rs(k) - cofrz(k) *(rw_p(k+1,iCell)-rw_p(k ,iCell))
! rtheta_pp(k,iCell) = ts(k) - rdzw(k)*(coftz(k+1,iCell)*rw_p(k+1,iCell) &
! -coftz(k ,iCell)*rw_p(k ,iCell))
! WCS_epssm
! wwAvg(k,iCell) = wwAvg(k,iCell) + 0.5*(1.0+epssm)*rw_p(k,iCell)
Let’s get rid of subroutine set_epssm(ewp,ewm,etp,etm,n,rdzw)
And subroutine calculate_computational_height(nVertLevels,rdzw,height_u_levels, height_w_levels)
MPAS version 8 to include: 1) an option to use full thermodynamic variables instead of perturbations from a reference profile and 2) a capability for specifying an epssm profile that is variable with height. These revisions include the changes suggested by Bill Skamarock in his review of the pull request.
Code added to ../src/core_atmosphere/mpas_atm_core.F to compute
vertical profiles of epssm that vary with height at the start of the
time integration. These profiles are determined by the parameters
config_min_coeff, config_max_coeff, config_transition_lower_bound,
and config_transition_upper_bound, and are specified in namelist.atmosphere.
Code added to allow thermodynamic variables to be represented either
as perurbations from a specified reference state or as the full
variables without a reference state. This option is controlled by the
logical config_reference_sounding as specified in namelist.init_atmosphere.
When configured using the full variables, the horizontal and vertical
pressure gradients computed on the large time steps are written in terms
of log(p).